Background Population prevalence estimates of electronic nicotine delivery system (ENDS) use range considerably based on the operational definition of ‘use’. Recently, we investigated the utility of ‘use frequency’ for restricting prevalence estimates to non-experimenters in adult populations. Results suggested that individuals reporting use on ≤5 days in the past 30 were likely to discontinue use, and should be excluded from estimates of population prevalence.

Objective This study investigated the predictive validity of ENDS use frequency as a measure for likelihood of continued use, and cigarette smoking abstinence.

Methods We recontacted smokers and recent quitters who participated in the random digit dial 2014 Minnesota Adult Tobacco Survey. At ∼1 year follow-up, we reassessed ENDS use and cigarette smoking among N=601 respondents.

Results Fewer than half of respondents who reported using ENDSs on 1–5 days in the past 30 at baseline reported any ENDS use 1 year later (27%, 95% CI (18% to 40%)). Conversely, more than half of respondents who reported daily use at baseline also reported subsequent use at follow-up (89%, 95% CI (78% to 100%)). The likelihood of subsequent ENDS use by respondents using ENDSs on more than 5 days but less than daily did not significantly differ from chance (37%, 95% CI (22% to 61%)).

Conclusions For adult population surveillance surveys, defining current use prevalence as ‘any use in the past 30 days’ includes many individuals who can be expected to discontinue use within 1 year. Until measures of ENDS use become standardised, researchers should choose definitions carefully because different definitions are likely to yield different results.